Forecasting Australian Consumer Sentiment

Forecasts are now available in the premium content directory of this website

Forecast updated 23 May 2005

Sentiment is forecast to remain at high levels for the next six months at least, although a bad GDP number for the March 2005 quarter could cause another temporary dip.

Forecasting Australian Consumer Sentiment

Charlie Nelson
director foreseechange
27 September 2002

Some time ago, my research showed that the Westpac/Melbourne Institute consumer sentiment index tells us nothing about the future of consumer spending.  This was reported in the Australian Financial Review of 20 February 2001 (page 29), Barrie Dunstan reported on research and forecasts that we published in January 2001.

Once we allow for interest rate movements, consumer confidence measures contain no information about the future of consumer spending, said Charlie Nelson.

After adjusting retail growth for interest rate movements, there is a significant but weak contemporaneous correlation with consumer sentiment.  If sentiment rises by 10% (from 100 to 110, for example) then retail turnover will increase by 0.4% in the same quarter, all other factors being equal.  Thus, the sentiment index contains a small amount of information about current consumer spending.  As the sentiment index is released in the month it is measured, several weeks before retail turnover data for that month is released, it has some small value in assessing current conditions.

On 17 January 2002, the Reserve Bank of Australia (RBA) released a research paper, dated November 2001, concerning both consumer and business sentiment surveys.  Their conclusion is that "... there is little evidence that the surveys tell us anything we didn't already know."

The RBA research approach was very interesting.  They first set out to find out what the index of sentiment is measuring – that is, what factors drive consumer sentiment.  They found that the official cash interest rate, ANZ job vacancies and the all ordinaries stock index were significant explanatory variables.  Then they looked at the residual (or unexplained) information in the consumer sentiment index and found that it had no significant correlation with economic variables such as retail turnover.  In the case of National Australia Bank’s business conditions and employment outlook surveys, the research found a small contribution to forecasting employment growth.

The RBA research used quarterly data because several economic measures, such as GDP, are only available quarterly.

We have undertaken a similar analysis of factors that drive the consumer sentiment index using monthly data.  We have found a slightly different set of significant explanatory variables that lead consumer sentiment– in other words, we can forecast the index of consumer sentiment.  We are conducting further research to improve our model.  In common with the RBA research, we have found that 1989 and 1992 are very difficult to explain, with sentiment falling a long way in 1989 and rising again in 1991 and 1992.  The 1989 drop may have been influenced by the actual or imminent collapse of several organisations such as Pyramid, Tricontinental, Estate Mortgage, National Safety Council, and the Bond and Skase empires.  The recovery in 1991 and 1992 may have been the correction to overly-pessimistic sentiment in 1989.

The editorial of the Australian Financial Review of 24 September 2002 has called for consideration of the private sector data collections, such as the Westpac/Melbourne Institute consumer sentiment index and the ANZ job ads surveys to be released simultaneously to all market participants, as is Australian Bureau of Statistics data.  This follows claims that some of the banks favoured clients are receiving a briefing before the data is released publicly.

If the data is privately funded, the banks and others who collect the data may do as they please with it.  Although is it privately funded?  Does the RBA or Treasury contribute to the funding?  If so, then perhaps there is a case for treating this data as “official” and releasing it simultaneously to the whole market.  Also, if the RBA or Treasury do contribute to the index of consumer sentiment then they should withdraw their financial support because both my research and the RBA’s show that the data has no predictive power and they should spend public money on something more useful.

Of course the Financial Review’s editorial should have acknowledged the very limited value of sentiment surveys.  Whatever the outcome of this storm in a teacup, you can now get a forecast of consumer sentiment before it is even measured.  We will release our forecast on our website (www.foreseechange.com/forecastingics.htm ) at the end of the month prior to measurement.

Back